The differences between isset
and empty
are subtle but important. They are most relevant when used alone. If you are checking that a variable exists and is a truethy value (e.g. any string that is not all spaces or 0s) you can use either interchangeably.
When to use isset
Use isset
when it's important to know if the variable has been defined and is not null:
if (isset($maybeExistsMaybeNull)) {
// variable defined and is not NULL
}
When to use !empty
Use !empty
when it's important to know if the variable has be defined and is truthy
if (!empty($mightBeEmpty)) {
// variable defined, and isn't "", " ", 0, "0" etc.
}
!empty
is a great shorthand for exists and is something.
When to use array_key_exists
Use array_key_exists
when it's important to know if the key exists and the value is of no importance:
if (array_key_exists('something', $array)) {
// $array['something'] exists, could be literally anything including null
}
When not to use isset
If your code looks like this:
if (isset($something) && $something) {
// code is shorter with !empty
}
When not to use !empty
If your code looks like this:
if (!empty($something) && $something === "") {
// you meant isset. this is unreachable.
}
Then you're writing code that can't be executed
Code that throws errors is error prone
Avoid writing code that issues notices/warnings that you are ignoring. For example in the question:
if((isset($_GET[example]))&&($_GET['example']=='somevalue')){ ... }
The first use of example is an undeclared constant. Or is it undeclared - what if you've got define('example', "foo");
somewhere else in the code.
if($_GET['example']=='somevalue'){ ... }
If the url doesn't contain ?example=..
that's going to issue a notice too.
Writing code without displaying errors means you can very easily miss mistakes like the first.
In context: isset and !empty are equivalent
For the example given, these two language constructs act exactly the same.
There is no case where one will act differently than the other, neither will issue a notice if the variable is undefined, and no measurable difference in performance between the two.